wx44wx/pipecat
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Refactoring how we are reading the image bytes inside the base llm.

Browse Source
This commit is contained in:
Filipi Fuchter
2025-09-23 08:05:15 -03:00
parent 7c569b3863
commit eaecefe675
1 changed files with 5 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
src/pipecat/services/openai/base_llm.py
View File

@@ -281,8 +281,10 @@ class BaseOpenAILLMService(LLMService):
# base64 encode any images # base64 encode any images
for message in messages: for message in messages:
if message.get("mime_type") == "image/jpeg": if message.get("mime_type") == "image/jpeg":
encoded_image = base64.b64encode(message["data"].getvalue()).decode("utf-8") # Avoid .getvalue() which makes a full copy of BytesIO
text = message["content"] raw_bytes = message["data"].read()
encoded_image = base64.b64encode(raw_bytes).decode("utf-8")
text = message.get("content", "")
message["content"] = [ message["content"] = [
{"type": "text", "text": text}, {"type": "text", "text": text},
{ {
@@ -290,6 +292,7 @@ class BaseOpenAILLMService(LLMService):
"image_url": {"url": f"data:image/jpeg;base64,{encoded_image}"}, "image_url": {"url": f"data:image/jpeg;base64,{encoded_image}"},
}, },
] ]
# Explicit cleanup
del message["data"] del message["data"]
del message["mime_type"] del message["mime_type"]